https://github.com/thirdgerb/compnt-rasa
component using rasa as nlu for commune/chatbot
https://github.com/thirdgerb/compnt-rasa
Last synced: 28 days ago
JSON representation
component using rasa as nlu for commune/chatbot
- Host: GitHub
- URL: https://github.com/thirdgerb/compnt-rasa
- Owner: thirdgerb
- License: mit
- Created: 2019-12-28T05:13:30.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2019-12-30T17:26:28.000Z (about 6 years ago)
- Last Synced: 2025-01-10T17:56:01.206Z (about 1 year ago)
- Language: PHP
- Size: 9.77 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
本项目是[commune/chatbot](https://github.com/thirdgerb/chatbot) 和 [commune/studio-hyperf](https://github.com/thirdgerb/studio-hyperf)的组件, 可将 [Rasa](https://rasa.io) 提供的 NLU 中间件引入项目.
```
composer require commune/compnt-rasa
```
## 注册 RasaComponent
CommuneChatbot 对 Rasa 服务做了基本的封装, 需要通过 ```Commune\Components\Rasa\RasaComponent``` 引入系统, 具体而言有三步:
1. 配置数组引入 ```Commune\Components\Rasa\RasaComponent```
1. ```NLUComponent::$nluSerivces``` 处, 注册```Commune\Components\Rasa\Services\RasaService```
1. 在配置数组 ```$chatbotConfig->host->sessionPipes``` 中引入中间件```Commune\Components\Rasa\RasaSessionPipe```
```php
// 机器人配置数组 ChatbotConfig
return [
...
'components' => [
...
Commune\Components\Rasa\RasaComponent::class => [
// rasa http api 地址
'server' => 'localhost:5005',
// 如果通过 jwt 验证的话
'jwt' => '',
// 意图的置信度, 高于 0.7 才加入 highlyPossibleIntent
'threshold' => 70,
// 语料库同步的目标文件位置
'output' => __DIR__ .'/resources/rasa.md',
// rasa domain.md 的文件位置
'domainOutput' => __DIR__ . '/resources/domain.yml',
// 使用 Guzzle 调用 Rasa api 时的配置
'clientConfig' => [
'timeout' => 0.3,
],
],
NLUComponent::class => [
'nluServices' => [
...
Commune\Components\Rasa\Services\RasaService::class,
]
],
],
...
'host' => [
...
'sessionPipes' => [
...
Commune\Components\Rasa\RasaSessionPipe::class,
]
...
],
]
```
## 同步语料库
Rasa 所需要的语料库, 格式在 [Rasa training data format](https://rasa.com/zh-cn/rasa/nlu/training-data-format/) 可以查看.
CommuneChatbot 的本地语料库设计参考了 Rasa 定义的格式.
使用 ```Commune\Components\Rasa\Services\RasaService``` 可以直接将本地语料库同步到 rasa 的配置文件中.
具体内容请查看 [开发手册](https://github.com/thirdgerb/chatbot-book).
```
CommuneChatbot 开发了一批用多轮对话管理机器人自己的工具.
请选择您要使用的开发工具: (1)
[1] nlu 语料库管理
> 1
选择功能 (1)
[1] 查看NLU命中的意图
[2] 同步语料库到NLU
[3] 同步语料库到本地存储
[4] 管理意图的语料样本
> 2
Commune\Components\Rasa\Services\RasaService : success
```